election_data_clean <- election_data |>
pivot_longer(cols ="BERDEAK-LOS VERDES":"COALICIÓN POR MELILLA", #Pivoting to make tidy data
names_to = "party",
values_to = "votes")
election_data_clean<-election_data_clean|>
mutate(date_elec = make_date(anno, mes)) |>
select(-tipo_eleccion, -codigo_distrito_electoral) |>
mutate(votes = ifelse(is.na(votes), 0, votes)) #changing all NA values to 0, as we expect that the data set misinterprets 0s as NAGroup presentation
Frederick, Nienke, Mika, Alex, Alicia
2024-01-24
Data cleaning
Cleaning data
Tidy data
Joining datasets
Election data
Abbreviation data
distinct_parties <- election_data_clean |>
distinct(party)
# Combine datasets as there are parties in election_data not present in
abbrev_test <- abbrev |>
full_join(distinct_parties, by = c("denominacion" = "party"))
#Creating a new column with the collected parties
abbrev_collect<-abbrev_test |>
mutate(party_collect = case_when(
str_detect(siglas, "PSOE") |
str_detect(siglas, "P.S.O.E") |
str_detect(siglas, "psoe") |
str_detect(denominacion, "PARTIDO SOCIALISTA") |
str_detect(denominacion, "PSOE") |
str_detect(denominacion, "P.S.O.E.") ~ "PSOE", # PARTIDO SOCIALISTA OBRERO ESPAÑOL
str_detect(denominacion, "CONVERGENCIA I UNIO") ~ "CU", # CONVERGÈNCIA I UNIÓ
str_detect(siglas, "PP") |
str_detect(denominacion, "PARTIDO POPULAR") ~ "PP", # PARTIDO POPULAR
str_detect(denominacion, "CIUDADANOS-") ~ "CS", #CIUDADANOS
str_detect(denominacion, "UNIDAS")|
str_detect(siglas, "PODEMOS")|
str_detect(denominacion, "VERDES")|
str_detect(denominacion, "PODEM")|
str_detect(denominacion, "EZKER") |
str_detect(siglas, "IU")~ "UP_IU", #UNIDAS PODEMOS - IU
str_detect(denominacion, "BILDU")|
str_detect(siglas, "EH")|
str_detect(denominacion, "SORTU")|
str_detect(denominacion, "ALKARTASUNA")|
str_detect(denominacion, "ARALAR") |
str_detect(denominacion, "ALTERNATIBA") ~ "EH_BILDU", # EH - BILDU
str_detect(denominacion, "VASCO") ~ "PNV", # PARTIDO NACIONALISTA VASCO
str_detect(siglas, "BNG") |
str_detect(denominacion, "GALLEG") |
str_detect(denominacion, "GALICIA") |
str_detect(denominacion, "GALEGUISTA")|
str_detect(denominacion, "GALEG") |
str_detect(denominacion, "GALIZ") ~ "BNG", # BLOQUE NACIONALISTA GALLEGO
str_detect(siglas, "VOX") ~ "VOX", # VOX
str_detect(siglas, "PAÍS") ~ "MP", # MÁS PAÍS
str_detect(siglas, "COMPROMÍS") ~ "CP", # COMPROMÍS
str_detect(siglas, "ERC") |
str_detect(denominacion, "CATALUNYA SÍ") ~ "ERC", # ESQUERRA REPUBLICANA DE CATALUNYA
TRUE ~ "OTHER"))|>
distinct(denominacion, .keep_all = TRUE)Survey data
survey_clean <- surveys |>
filter(year(date_elec) >= 2018) |> #Filtering only years >2018
filter(exit_poll == FALSE) |> # Only keeping non exit polls
filter(size >750| size != NA) |> #Only keeping cases with a size ,larger than 750 or are bot NA
mutate(field_days = field_date_to - field_date_from) |> #New column to count the number of field days
relocate(field_days, .after = field_date_to) |> #Relocating the field_days column
filter(field_days>0) |> #Filtering cases with field days >0
pivot_longer(cols ="UCD":"EV", #Pivoting to make tidy data
names_to = "party",
values_to = "vote_int") |>
mutate(vote_int = ifelse(is.na(vote_int), 0, vote_int))Joining tables
Joining election data with abbrev_collect
Joining survey data with abbrev_collect
#Joining the abbrev_collect table with the survey table and adding the abbreviations that are not in the abbrev table to the column: party_collect
survey_clean_collect<-survey_clean |>
left_join(abbrev_collect, by = c("party" = "siglas")) |>
mutate(party_collect =
case_when(party_collect == "PSOE" ~ "PSOE",
party_collect == "CS" ~ "CS",
party_collect == "UP_IU" ~ "UP_IU",
party_collect == "EH_BILDU" ~ "EH_BILDU",
party_collect == "PP" ~ "PP",
party_collect == "PNV" ~"PNV",
party_collect == "BNG" ~ "BNG",
party_collect == "VOX" ~ "VOX",
party_collect == "MP" ~ "MP",
party_collect == "CP" ~ "CP",
party_collect == "ERC" ~ "ERC",
party_collect == "CU" ~ "CU",
TRUE ~ party)) |>
select(-party)
#Reducing the data set by taking out unsuseful variables
survey_clean_collect_red <- survey_clean_collect |>
select(-field_date_from, -field_date_to, -exit_poll) |>
drop_na(vote_int)
# removing siglas not to confuse with party_collect
abbrev_collect<- abbrev_collect |>
select(-siglas)Creating IDs
Creating an ID to match the election_data table
Q1
How is the vote of national parties (PSOE, PP, VOX, CS, MP, UP - IU) distributed against regional or nationalist parties?
national_parties <- c("PSOE", "PP", "VOX", "CS", "MP", "UP_IU")
regional_nationalist_parties <- c("OTHER", "ERC", "CP", "BNG", "CU", "PNV", "EH_BILDU" )
#data set I need for my question
election_data_clean_1 <- election_data_clean |>
select(-(anno:votos_candidaturas))
#slicing the sample to make the code and then will run it when its all done
#election_data_sample <- election_data_clean_1 |>
#slice_sample(prop =0.10)
election_data_q1 <- election_data_clean_1 |>
group_by(date_elec, party_collect) |>
mutate(party_type =
case_when(party_collect %in% c("PSOE", "PP", "VOX", "CS", "MP", "UP_IU") ~ "national",
TRUE ~ "regional/nationalist")) |>
mutate(sum_votes = sum(votes)) |>
ungroup()
#Summing the total votes per party per election
election_data_q1 <- election_data_q1 |>
group_by(date_elec, party_type) |>
mutate(votes_elec_party = sum(votes)) |>
ungroup()
#Summing the total votes voted per election
election_data_q1 <- election_data_q1 |>
group_by(date_elec) |>
mutate(votes_elec_total = sum(votes)) |>
ungroup()
#Computing a percentage of how many votes a party received compared to the total votes voted
election_data_q1 <- election_data_q1 |>
mutate(votes_elec_party_perc = (votes_elec_party/votes_elec_total)*100)
#creating new dataset to visualize the relationship between the share of national parties and regional/ nationalist parties
graph_data <- election_data_q1 |>
group_by(date_elec) |>
distinct(party_type, votes_elec_party_perc) Q1 Visualisation
p1<-ggplot(data = graph_data, aes(x = date_elec, y = votes_elec_party_perc, color = party_type, group = party_type)) +
geom_line() +
geom_text(aes(label = sprintf("%.1f%%", votes_elec_party_perc)), vjust = -0.5, hjust = 0.5, size = 3, color = "black") +
labs(title = "Vote Share of National and Regional Parties\nin Elections between 2008 and 2019",
y = "",
x = "Date of Election") +
scale_y_continuous(breaks = seq(0, 100, by = 10))+
scale_x_date(date_breaks = "1 year", date_labels = "%Y")+
theme_classic() +
theme(legend.position = "none")Q1 Visualisation
Q2
Which party was the winner in the municipalities with more than 100,000 habitants (census) in each of the elections?
election_data_clean_2 <-election_data_clean |>
select(-(anno:vuelta)) |>
mutate(codigo_ccaa_pro_muni = paste(codigo_ccaa, codigo_provincia, codigo_municipio, sep = "-")) |> #code to track each by municipality
mutate(total_votes_municipality = votos_blancos+ votos_nulos + votos_candidaturas) |> #total votes casted in each municipality for every row
group_by(date_elec, codigo_ccaa_pro_muni) |>
mutate(votes_perc_party_municipality = (votes/total_votes_municipality)*100) |> #percentage of votes that a party got from the total amount of votes cast in each municipality
ungroup()
q2_final_result <- election_data_clean_2 |>
filter(censo > 100000) |>
group_by(date_elec, codigo_ccaa_pro_muni) |>
arrange(desc(votes_perc_party_municipality)) |>
slice(1) |>
ungroup() Q2 Visualisation
p2 <-
ggplot(q2_final_result,
aes(x =party_collect))+
geom_bar(stat = "count", fill = "black") +
geom_text(stat = "count", aes(label = after_stat(count)), vjust = -0.5, size = 3) +
labs(title = "Election Winners in Municipalities with more than 100,000 inhabitants in each election",
x = "Party",
y = "Count") +
guides(fill = guide_legend(title = "Party")) +
theme_classic() +
facet_wrap(~date_elec, scales = "free_x")Q2 Visualisation
Q3
Which party was the second when the first was the PSOE? And when the first was the PP?
election_data_q3 <- election_data_clean_2 |>
left_join(cod_mun, by = c("codigo_ccaa_pro_muni" = "cod_mun")) |>
filter(censo > 100000) |>
select(date_elec, party, party_collect, codigo_ccaa_pro_muni, municipio, votes_perc_party_municipality) |>
group_by(date_elec, codigo_ccaa_pro_muni) |>
arrange(desc(votes_perc_party_municipality)) |>
mutate(rank = row_number()) |>
ungroup()
# Identify dates and municipalities where PSOE won
psoe_wins <- election_data_q3 |>
filter(party_collect == "PSOE", rank == 1) |>
select(date_elec, codigo_ccaa_pro_muni)
# Filter for municipalities where PSOE won and find the parties coming in second place
parties_2nd_place_after_PSOE <- election_data_q3 |>
semi_join(psoe_wins, by = c("date_elec", "codigo_ccaa_pro_muni")) |>
filter(rank == 2) |>
select(date_elec, municipio, party, party_collect)
# Identify dates and municipalities where PP won
PP_wins <- election_data_q3 |>
filter(party_collect == "PP", rank == 1) |>
select(date_elec, codigo_ccaa_pro_muni)
# Filter for municipalities where PP won and find the parties coming in second place
parties_2nd_place_after_PP <- election_data_q3 |>
semi_join(PP_wins, by = c("date_elec", "codigo_ccaa_pro_muni")) |>
filter(rank == 2) |>
select(date_elec, municipio, party, party_collect) Q3 Visualisation I
This barplot illustrates the second place finisher for municipalities where the PSOE won.
p3 <-
ggplot(parties_2nd_place_after_PSOE,
aes(x =party_collect))+
geom_bar(stat = "count", fill = "black") +
geom_text(stat = "count", aes(label = after_stat(count)), vjust = -0.5, size = 3) +
labs(title = "Second placed parties if PSOE wins a municipality",
x = "Party",
y = "Count") +
guides(fill = guide_legend(title = "Party")) +
theme_classic()Q3 Visualisation I
Q3 Visualisation II
This barplot illustrates the second place finisher for municipalities where the PP won.
p4 <-
ggplot(parties_2nd_place_after_PP,
aes(x =party_collect))+
geom_bar(stat = "count", fill = "black") +
geom_text(stat = "count", aes(label = after_stat(count)), vjust = -0.5, size = 3) +
labs(title = "Second placed parties if PP wins a municipality",
x = "Party",
y = "Count") +
guides(fill = guide_legend(title = "Party")) +
theme_classic()Q3 Visualisation II
Q3 Answer
We now only read out the party that has come in second the most often when the PSOE has won using the glue package.
The party that has come in 2nd most often when the PSOE has won a municipality is PP.The party that has come in 2nd most often when the PP has won a municipality is PSOE.Q4
Who benefits from low turnout?
#Create turnout column
election_data_turnout <- election_data_clean |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
group_by(municipio_id, date_elec) |>
mutate(
turnout_pcg = round((((votos_candidaturas+votos_nulos+votos_blancos)/censo)*100), digits = 2)) |>
ungroup() |>
group_by(municipio_id) |>
mutate(vote_share = (votes / (votos_candidaturas + votos_blancos+votos_nulos))*100) |>
ungroup()
#Low turnout dataset
low_turnout_winners <- election_data_turnout |>
filter(turnout_pcg < 60) |>
dplyr::slice_max(n = 1, vote_share, by = c(municipio_id)) |>
select(vote_share, party, party_collect)Q4 Visualisation
Historically, since 2008, in low turnout elections, smaller parties have had a higher chance of winning the election.
p5 <-
ggplot(low_turnout_winners, aes(x = party_collect)) +
geom_bar(stat = "count") +
geom_text(stat = "count", aes(label = after_stat(count)), vjust = -0.5, size = 3) +
labs(title = "Number of Municipalities Won by Party in Low Turnout Elections",
caption = "Low turnout defined as any election with a turnout less than 60% ",
x = "Party",
y = "") +
scale_y_continuous(breaks = seq(0, 55000, by = 10000)) +
theme_classic()Q4 Visualisation
Q5
How to analyze the relationship between census and vote? Is it true that certain parties win in rural areas?
#Subset with rural municipalities and their winners
election_winners_rural<- election_data_clean |>
filter(censo < 30000, votes > 0) |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
group_by(date_elec, municipio_id) |>
mutate(vote_share = (votes / (votos_candidaturas + votos_blancos + votos_nulos))*100) |>
ungroup() |>
dplyr::slice_max(n = 1, vote_share, by = c(municipio_id)) |>
select(vote_share, party, party_collect)Q5 Visualisation
As may be expected, the PP is most successful in elections in rural areas.
#Plot the count of each party winning rural areas
p6 <-
ggplot(election_winners_rural, aes(x = party_collect)) +
geom_bar(stat = "count") +
geom_text(stat = "count", aes(label = after_stat(count)), vjust = -0.5, size = 3) +
labs(title = "Winning Parties in Rural Areas",
caption = "Rural area defined as municipality with less than 30,000 inhabitants",
x = "Party",
y = "Count")+
theme_classic()Q5 Visualisation
Q6
How to calibrate the error of the polls (remember that the polls are voting intentions at national level)?
Q6 II
election_data_clean_red <- election_data_clean |>
select(-anno, -mes, -vuelta,-participacion_1, -participacion_2, -votos_candidaturas) |>
mutate(codigo_ccaa_pro_muni = paste(codigo_ccaa, codigo_provincia, codigo_municipio, sep = "-")) |>
filter(year(date_elec) >= 2018)
#Summing the total votes per party per election
election_data_clean_red<-election_data_clean_red |>
group_by(date_elec, party_collect) |>
mutate(votes_elec_party = sum(votes)) |>
ungroup()
#Summing the total votes voted per election
election_data_clean_red<-election_data_clean_red |>
group_by(codigo_ccaa_pro_muni, date_elec) |>
mutate(votes_total_blancos_nulos = votos_blancos + votos_nulos) |>
ungroup() |>
group_by(date_elec) |>
mutate(votes_total_blancos_nulos = sum (votes_total_blancos_nulos)) |>
mutate(votes_elec = sum(votes)) |>
ungroup() |>
mutate(votes_elec_total = votes_total_blancos_nulos + votes_elec)
#Computing a percentage of how many votes a party received compared to the total votes voted
election_data_clean_red<-election_data_clean_red |>
mutate(votes_elec_party_perc = (votes_elec_party/votes_elec_total)*100)
election_data_clean_red_7 <- election_data_clean_red |>
select(id_join, votes_elec_total, codigo_ccaa, date_elec, votes_elec_party_perc, party_collect) |>
distinct(id_join, date_elec, .keep_all = TRUE)
survey_clean_collect_red_7 <- survey_clean_collect_red |>
select(id_join, vote_int, id_pollster)
# Joining survey_clean with election_data_clean using the id_join from both tables
survey_elec <- election_data_clean_red_7 |>
left_join(survey_clean_collect_red_7, by = c("id_join" = "id_join"), relationship = "many-to-many")
#subtracting the voting intentions (in percentage) of the percenatage of votes gained per per party per election.
#As the difference can be both positive and negative, the differences in votes are squared to account for the negatives
survey_elec<-survey_elec |>
mutate(diff_vote_int = votes_elec_party_perc - vote_int) |>
mutate(diff_vote_int_2 =diff_vote_int^2)
#Creating the average difference between percentage of votes gained and voting intentions per election
#Taking the squareroot to return the variable in to the orginal units
survey_elec<-survey_elec |>
group_by(date_elec) |>
mutate(avg_diff_vote_int = sqrt(mean(diff_vote_int_2, na.rm = TRUE))) |>
ungroup()Q7
In which election were the polls most wrong?
Q7 Visualisation
Q7 Visualisation
Q8
How were the polls wrong in national parties (PSOE, PP, VOX, CS, MP, UP - IU)?
#Computing the average polling error per party per election
survey_elec_national<-survey_elec |>
filter(party_collect == "PSOE"|party_collect == "PP"|party_collect =="VOX"|party_collect =="CS "|party_collect =="MP"|party_collect =="UP_IU") |>
group_by(date_elec, party_collect) |>
mutate(avg_diff_poll = sqrt(mean(diff_vote_int_2, na.rm = TRUE)) ) |>
ungroup() Q8 II
#Obtaining the results
survey_elec_national |>
distinct(avg_diff_poll, date_elec, party_collect) |>
arrange(date_elec)# A tibble: 10 × 3
avg_diff_poll date_elec party_collect
<dbl> <date> <chr>
1 1.55 2019-04-01 UP_IU
2 18.7 2019-04-01 PP
3 25.3 2019-04-01 PSOE
4 9.91 2019-04-01 VOX
5 0 2019-04-01 MP
6 1.32 2019-11-01 UP_IU
7 18.6 2019-11-01 PP
8 25.4 2019-11-01 PSOE
9 9.83 2019-11-01 VOX
10 3.39 2019-11-01 MP Q8 Visualisation
Q8 Visualisation
Q9
Which polling houses got it right the most and which ones deviated the most from the results?
#Computing the average polling error per polling house per election and obtaining the minimum and maximum polling with their corresponding pollhouses per election
survey_elec<-survey_elec |>
group_by(id_pollster, date_elec) |>
mutate(avg_diff_pollhouse = sqrt(mean(diff_vote_int_2, na.rm = TRUE))) |>
ungroup() |>
drop_na(avg_diff_pollhouse)
#Obtaining the result for the maximum polling error
result_max<-survey_elec|>
group_by(date_elec) |>
filter(avg_diff_pollhouse == max(avg_diff_pollhouse)) |>
distinct(avg_diff_pollhouse, id_pollster, date_elec) Q9
# A tibble: 2 × 3
# Groups: date_elec [2]
avg_diff_pollhouse id_pollster date_elec
<dbl> <chr> <date>
1 9.74 pollster-49 2019-04-01
2 10.0 pollster-76 2019-11-01Q9 II
#Obtaining the result for the minimum polling error
result_min<-survey_elec|>
group_by(date_elec) |>
filter(avg_diff_pollhouse == min(avg_diff_pollhouse)) |>
distinct(avg_diff_pollhouse, id_pollster, date_elec)
result_min# A tibble: 2 × 3
# Groups: date_elec [2]
avg_diff_pollhouse id_pollster date_elec
<dbl> <chr> <date>
1 8.72 pollster-56 2019-04-01
2 8.42 pollster-49 2019-11-01Additional insights
Voter turnout: Regions
# Voter turnout regions
election_data_clean_lollipop <-
election_data_clean |>
filter(date_elec >= "2019-04-01") |>
mutate(votes_casted = votos_candidaturas + votos_blancos + votos_nulos) |>
select(-votos_blancos, -votos_candidaturas, -votos_nulos) |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
distinct(municipio_id, mes, .keep_all = T) |>
select(municipio_id, codigo_ccaa, votes_casted, censo, mes) |>
group_by(codigo_ccaa, mes) |>
summarise(votes_sum = sum(votes_casted),
population_sum = sum(censo) ) |>
ungroup() |>
mutate(voter_turnout = (votes_sum / population_sum) * 100) Voter turnout: Spain
# Voter turnout Spain
election_data_clean_lollipop_avg <-
election_data_clean |>
filter(date_elec >= "2019-04-01") |>
mutate(votes_casted = votos_candidaturas + votos_blancos + votos_nulos) |>
select(-votos_blancos, -votos_candidaturas, -votos_nulos) |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
distinct(municipio_id, mes, .keep_all = T) |>
select(municipio_id, codigo_ccaa, votes_casted, censo, mes) |>
group_by(mes) |>
summarise(votes_sum = sum(votes_casted),
population_sum = sum(censo) ) |>
mutate(voter_turnout = (votes_sum / population_sum) * 100)
election_data_clean_lollipop_avg$codigo_ccaa <- "00"
election_data_clean_lollipop <- bind_rows(election_data_clean_lollipop, election_data_clean_lollipop_avg)Dumbell graph data preperation I
# ordering codigo variable for subsequent plot
max_turnout_mes04 <- election_data_clean_lollipop |>
filter(mes == "04") |>
group_by(codigo_ccaa) |>
summarize(max_turnout = max(voter_turnout))
# Order codigo_ccaa based on max_turnout_mes11
order_by_max_turnout <- max_turnout_mes04 |>
arrange(max_turnout) |>
pull(codigo_ccaa)
# Apply the order to the codigo_ccaa factor
election_data_clean_lollipop$codigo_ccaa <- factor(
election_data_clean_lollipop$codigo_ccaa,
levels = order_by_max_turnout
)Dumbell graph data preperation II
# for segments
data_00 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "00")
data_01 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "01")
data_02 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "02")
data_03 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "03")
data_04 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "04")
data_05 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "05")
data_06 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "06")
data_07 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "07")
data_08 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "08")
data_09 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "09")
data_10 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "10")
data_11 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "11")
data_12 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "12")
data_13 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "13")
data_14 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "14")
data_15 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "15")
data_16 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "16")
data_17 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "17")
data_18 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "18")
data_19 <- election_data_clean_lollipop |>
filter(codigo_ccaa == "19")
election_data_clean_lollipop_dif <- election_data_clean_lollipop |>
group_by(codigo_ccaa) |>
mutate(difference = voter_turnout - lag(voter_turnout, default = voter_turnout[1])) |>
ungroup() |>
filter(mes == "11")Dumbell graph - voter turnout differences
p9 <-
ggplot(election_data_clean_lollipop, aes(x = voter_turnout, y = codigo_ccaa, group= mes), size = 2) +
geom_point(aes(color = mes)) +
labs(title = "Difference in Voter Turnout\nby Region (April vs. November 2019)",
x = "",
y = "") +
guides(col= guide_legend(title= "Election")) +
scale_color_manual(labels = c("April 2019", "November 2019"), values = c("blue", "red")) +
scale_x_continuous(labels = scales::percent_format(scale = 1),
limits = c(45, NA),
breaks = seq(45, 80, by = 5),) +
geom_segment(data = data_00, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_01, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_02, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_03, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_04, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_05, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_06, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_07, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_08, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_09, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_10, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_11, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_12, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_13, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_14, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_15, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_16, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_17, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_18, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_segment(data = data_19, aes(xend = max(voter_turnout), yend = codigo_ccaa), color = "black") +
geom_text(data = election_data_clean_lollipop_dif,
aes(label = sprintf("%.2f%%", difference),
x = voter_turnout,
y = codigo_ccaa),
vjust = 0.4,
hjust = 1.1,
size = 3,
color = "black") +
theme_minimal() +
scale_y_discrete(labels=c("00" = "Average in Spain",
"01"="Andalucia",
"02"="Aragon",
"03"="Asturias",
"04"="Baleares",
"05"="Cantabria",
"06"="Castilla La Mancha",
"07"="Castilla Leon",
"08"="Cataluna",
"09"="Canarias",
"10"="Extremadura",
"11"="Galicia",
"12"="Madrid",
"13"="Murcia",
"14"="Navarra",
"15"="Comunidad Valenciana",
"16"="Pais Vasco",
"17"="La Rioja",
"18"="Ceuta",
"19"="Melilla")) +
theme(plot.margin = margin(5, 10, 5, -8))Dumbell graph
Election Winners in November 2019
Data preperation.
election_data_clean_red_winner_2019_b <-
election_data_clean |>
dplyr::select(-anno, -mes, -vuelta,-participacion_1, -participacion_2) |>
filter(date_elec == "2019-11-01") |>
drop_na(votes) |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
group_by(date_elec, municipio_id) |>
mutate(vote_share = (votes / (votos_candidaturas + votos_blancos + votos_nulos))*100) |>
ungroup() |>
dplyr::slice_max(n = 1, vote_share, by = c(municipio_id)) |>
dplyr::select(municipio_id, vote_share, party_collect, party)Election Winners in November 2019 II
Merging with geospatial data.
esp_can <- esp_get_country()
can_prov <- esp_get_can_provinces()
can_box <- esp_get_can_box()
munic <- esp_get_munic()
# creating province_municipality code
munic <-
munic |>
mutate(municipio_id = paste(cpro, cmun, sep = "-"))
# merging based on province_municipality code
elec_winner_munic_2019b <- munic |>
left_join(election_data_clean_red_winner_2019_b, by = c("municipio_id" = "municipio_id")) Map Election Winners in November 2019
# party colors
party_colors <- c("UP_IU"="#8968CD",
"PSOE"="#CD2626",
"PP"="#0055A4",
"VOX"="#5ABF35",
"BAI"="#EECFA1",
"OTHER" = "#999999",
"PNV" = "#008341",
"EH_BILDU" = "#00D0B6",
"CS" = "#FF5001",
"ERC" = "#F0D85D",
"BNG" = "#77B4E0",
"MP" = "#02988B")
p10 <-
ggplot(esp_can) +
geom_sf() +
geom_sf(data = can_prov) +
geom_sf(data = can_box) +
geom_sf(data = elec_winner_munic_2019b, aes(fill = party_collect), color = "black", size = 1) +
scale_fill_manual(name = "Party", values = party_colors) +
theme_void() +
labs(title =" Election Winners by Municipality in November 2019") +
theme(legend.position = c(0.18, 0.55),
legend.justification = "center",
legend.background = element_rect(fill = NULL, colour = "black"),
legend.margin = margin(6, 6, 6, 6))+
guides(fill = guide_legend(ncol = 2)) +
coord_sf()Map Election Winners in November 2019
All elections since 2008
Data preperation.
election_winners <-
election_data_clean |>
dplyr::select(-mes, -vuelta,-participacion_1, -participacion_2) |>
drop_na(votes) |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
group_by(date_elec, municipio_id) |>
mutate(vote_share = (votes / (votos_candidaturas + votos_blancos + votos_nulos))*100) |>
ungroup() |>
dplyr::slice_max(n = 1, vote_share, by = c(municipio_id, date_elec)) |>
dplyr::select(municipio_id, vote_share, party_collect, party, anno, date_elec)
elec_winners_all_time <- munic |>
left_join(election_winners, by = c("municipio_id" = "municipio_id"))Mapping all elections since 2008
election_names <-
as_labeller(c(
'2008-03-01'="March 2008",
'2011-11-01'="November 2011",
'2015-12-01'="December 2015",
'2016-06-01'="June 2016",
'2019-04-01'="April 2019",
'2019-11-01'="November 2019"
))
p11 <-
ggplot(esp_can) +
geom_sf() +
geom_sf(data = can_prov) +
geom_sf(data = can_box) +
geom_sf(data = elec_winners_all_time, aes(fill = party_collect), color = "black", size = 1) +
scale_fill_manual(name = "Party", values = party_colors) +
theme_void() +
labs(title =" Election Winners by Municipality since 2008") +
theme(legend.position = "bottom",
legend.margin = margin(6, 6, 6, 6))+
guides(fill = guide_legend(nrow = 1)) +
coord_sf() +
facet_wrap(~date_elec, labeller = as_labeller(election_names))Mapping all elections since 2008
The 2019 Elections
Data preperation.
election_winners_2019 <-
election_data_clean |>
filter(anno == "2019")|>
dplyr::select(-mes, -vuelta,-participacion_1, -participacion_2) |>
drop_na(votes) |>
mutate(municipio_id = paste(codigo_provincia, codigo_municipio, sep = "-")) |>
group_by(date_elec, municipio_id) |>
mutate(vote_share = (votes / (votos_candidaturas + votos_blancos + votos_nulos))*100) |>
ungroup() |>
dplyr::slice_max(n = 1, vote_share, by = c(municipio_id, date_elec)) |>
dplyr::select(municipio_id, vote_share, party_collect, party, anno, date_elec)
elec_winners_2019 <-
munic |>
left_join(election_winners_2019, by = c("municipio_id" = "municipio_id"))Mapping the 2019 Elections
election_names <- as_labeller(c(
'2019-04-01'="April 2019",
'2019-11-01'="November 2019"
))
p12 <-
ggplot(esp_can) +
geom_sf() +
geom_sf(data = can_prov) +
geom_sf(data = can_box) +
geom_sf(data = elec_winners_2019, aes(fill = party_collect), color = "black", size = 1) +
scale_fill_manual(name = "Party", values = party_colors) +
theme_void() +
labs(title =" Election Winners by Municipality in 2019 elections") +
theme(legend.position = "bottom",
legend.margin = margin(6, 6, 6, 6),
legend.justification = "center")+
guides(fill = guide_legend(nrow = 1)) +
coord_sf() +
facet_wrap(~date_elec, labeller = as_labeller(election_names))Mapping the 2019 Elections
Vote distribution
election_data_mika <-
election_data_clean |>
select(c(party_collect, votes, date_elec))
election_data_mika$party_collect<-
factor(election_data_mika$party_collect)
election_data_mika<- election_data_mika |>
group_by(date_elec) |>
mutate(TOTALvotos_poreleccion = sum(votes))|>
ungroup()|>
summarise(votos_partidos = sum(votes),
.by = c(party_collect,
date_elec,
TOTALvotos_poreleccion)) |>
na.omit()
election_data_mika<- election_data_mika |>
mutate("vote_share" = round( votos_partidos/TOTALvotos_poreleccion*100, 2))
election_data_mika <- election_data_mika |>
mutate(convocatorias = format_ISO8601(date_elec,
precision = "ym"))Slopegraph
Slopegraph
Network analysis
Network analysis plot
p14 <-
plot(network,
main="Pollster & Media network",
layout= layout_nicely,
vertex.size=deg*9,
vertex.label.color = "black",
vertex.label.cex = 0.6,
vertex.label.degree = -pi/2,
edge.arrow.size = 0,
edge.arrow.width = 0,
edge.color = "black")
NULL